Automation Isn’t The Job Killer It’s Cracked Up To Be
CNN says automation is to blame for America’s manufacturing job losses; and what’s worse, there’s more to come—and no, it has absolutely nothing to do with bad trade deals like NAFTA, nor China’s predatory trading strategy.
Nope. Nothing to see here people. Move along.
Automation is bad, and your jobs are never coming back. This view was echoed recently by the Guardian too.
Of course, this is all just fake news—CNN has to earn its reputation somehow.
In reality, automation is only half the story: if you look at its impact relative to output, you’ll see that automation doesn’t really impact employment rates—unless things are woefully out of balance (they are, which explains the media hysteria, but I’ll get to that later).
In fact, automation, and better technology, is the only way to grow the economy in the long run.
Let’s look at the logic.
The Impact Of Automation On Employment
What is the actual impact of automation on unemployment?
At the most bare bones level, employment is predicated on the interplay between productivity and output—higher productivity increases unemployment, while higher output decreases unemployment; if they both increase at the same rate, then unemployment isn’t effected.
Let’s drop the economic jargon.
All that means is that the number of jobs is determined by 2 factors: (1) how fast stuff is made (productivity), and (2) how much stuff is made (output).
On the one hand, the faster we make stuff, the fewer workers we need to make it; on the other, if we make more stuff, we need more people to pitch in—assuming all things remain equal.
This means that the overall number of workers will only change if the 2 factors are imbalanced: if we get better at making stuff faster than we make more stuff (if we automate faster than we grow the economy), then we will need fewer workers, and there will be more unemployed people. And vice versa.
Essentially, we can only understand the impact of automation on employment with reference to output growth.
That’s how the logic works. Now let’s flesh it out with an example.
An Example Showing The Effect Of Automation On Unemployment
Pretend you’re the proud owner of Acme Automobiles, America’s most assiduous automaker.
Acme makes 1,000 awesome vehicles every year at their factory in Alhambra, Arizona. They employ 100 workers at said factory.
Luck strikes: 2016 was a banner year—Acme sold out of their premier Avalon Crossover by October.
For 2017 they have big plans: they’re going to make 100 extra Avalons in Alhambra.
This means they need to hire 10 more people, since 10% more output means 10% more jobs (assume for the sake of argument that there are no diminishing or increasing returns or anything else, let’s keep it simple).
In this case, increasing output created jobs.
Now instead of good luck, let’s pretend 2016 was a totally normal year: you sold 1,000 vehicles and don’t think you could sell more in 2017.
Being the prudent investor that you are, you decide that if you can’t make more money by selling extra Avalons, you’ll get Acme to invest in some robot welders—their assembly line now looks like this:
Aside from being cool, these make your factory more productive (efficient)—you only need 90 workers to make 1,000 cars. So you let 10 workers go (with a generous severance of course, you’re not an animal).
In this case, there was a loss of jobs due to automation—robot welders replaced humans.
Finally, let’s combine the two factors.
Pretend 2016 was a great year, and you decide to make 100 extra vehicles. Not only that, but because you’re a savvy businessman, you also decide to invest in the robot welders.
On the one hand, you need more workers to make more vehicles; on the other, you need fewer workers because some jobs were eliminated by technology.
But overall, the number of workers doesn’t change—you need the same number of workers, but now you can make more vehicles.
This is how better technology and economic growth create prosperity (wealth per person).
As you can see, so long as output grows just as fast as productivity, there won’t be any jobs lost to automation.
The theory’s sound, but does it match reality?
Are Jobs Lost To Automation? A Historical Context.
Both history and contemporary data show, beyond the shadow of a doubt, that there is no net loss of jobs due to automation (in the long run).
In by book Bobbins, Not Gold (which you need to read) I talk about how even in the 1100’s AD Europe was already beginning to automate. For example, textile weavers in Flanders invented elaborate machinery to full and weave cloth—this didn’t cause Flanders to have a bunch of unemployed weavers.
It made them Europe’s industrial heartland, and most wealthy, and highly urbanized region.
The same thing happened in the aftermath of the Black Death, which struck Europe in the 1350’s (depending upon where you were). Lots of people died, which meant that people needed to get creative if they were to continue building their civilization.
During this time Europe underwent the “First Industrial Revolution” (as I, and other Medievalists refer to it), which saw the widespread adoption of paddle-powered engines (river mills) to automatically grind grain, full cloth, and chop wood.
This, of course, culminated with the Renaissance.
There are countless examples of similar situations, including the Industrial Revolution in Great Britain during the 19th century (or perhaps more properly, the Third Industrial Revolution), where technology grew rapidly, but did not cause systemic unemployment.
The common thread of economic history is that better technology, vis-a-vis automation, has invariably made the West more prosperous.
Furthermore, there has been no long term loss of jobs due to automation—unemployment is temporary so long as economic growth continues (which is why the Luddites disappeared so quickly).
Debunking The “Loss Of Jobs Due To Automation” Theory Once And For All
Who do we blame?
Well, we can’t blame the robots.
As it turns out, the rate of productivity growth (a proxy for automation) hasn’t changed all that much since the 1800s (it’s actually declined from its peak in the 19th century).
Between 1950 and 1979, manufacturing employment increased because output grew faster than productivity. This was mirrored in the economy more broadly: unemployment was relatively low, and the economy grew faster than technology—there were lots of jobs, wages were on the rise.
But this changed in the 1980’s, when manufacturing output began to slow relative to productivity (we kept making more, but not as fast as we used to).
For example, between 1989 and 2000, manufacturing output grew by 3.7% per year on average, but productivity grew by 4.1%. This led to an average employment decline of 0.4% per year.
Since 2000, output has grown by only 0.4% per year on average, while productivity continues to increase by 3.7%.
This is why people (falsely) claim that automation is a job killer: they look at one side of the equation without looking at the other—if productivity increased like it used to (and did historically), then there wouldn’t be a problem, and we wouldn’t be having this conversation.
Both the logic and the historical evidence clearly show that there is no unemployment due to automation, in the long run.
Yes, there are some jobs eliminated by technology, but if output continues to rise, they are soon absorbed.
So, what changed? Why did output growth slow so dramatically?
America’s Job Loss: Automation vs Outsourcing
The reason that productivity and output were linked historically is because of a concept called capital immobility, which meant that capital (like factories) couldn’t be relocated to a different country.
Basically, stuff needed to be made pretty close to where it was consumed.
Capital immobility, as it turns out, is one of the underlying premises of comparative advantage, and global free trade. Without it, offshoring can (and does) destroy some national economies at the expense of others.
Not so coincidentally, America’s industrial output slowed when we started offshore outsourcing our production to countries like China and Mexico, rather than expanding capacity via automation in Michigan or Illinois.
Instead of making stuff in America, we made it abroad and imported it.
This trend is reflected in something called the goods trade deficit (which is the balance of imports vs exports).
Basically, everything we import replaces something we would otherwise need to make.
For example, if America needs 10 million pocket protectors, we can either make them, import them, or make some and import some.
If we imported 8 million pocket protectors, we would only need to make 2 million (the imports replace our production, not our consumption of pocket protectors)—imports replace our production, not our consumption of stuff.
Therefore, the deficit is the value of America’s offshored production.
“Automation Job Loss” Is A Myth To Cover For Bad Trade Policy
This is the million dollar question: how many “jobs lost to automation” were actually lost because of offshore outsourcing?
A good place to start is by looking at how many manufacturing jobs we’ve lost: 7 million. But we can do better than that.
American manufacturing contributes $2.2 trillion dollars to our economy. Meanwhile, our trade deficit is almost $750 billion a year.
Since 78% of our trade deficit is in manufactured goods, this means that we’ve offshored $573 billion worth of production. That’s one-third of our manufacturing industry.
Since manufacturing employs 12.3 million Americans, we know that roughly 4 million more are displaced by imports.
Of course, labor-intensive industries are the first to go (since labor will be a proportionately higher share of costs), which is why the numbers don’t match up exactly.
But let’s continue on with 4 million. There’s more to it.
Manufacturing brings wealth into a region (just like an oilfield or a mine, it produces something of value for export), and therefore supports local services and supply chains. For example, Acme Automobile’s factory supports hairdressers and accountants, but not the other way around.
Manufacturing (like other anchor industries) has a “job multiplier”, the value of which has been studied extensively.
As it turns out, each manufacturing job usually supports 1.58 other service jobs. This means that since 4 million manufacturing jobs are displaced by imports, then about 6 million service jobs were also lost.
According to this method, the trade deficit costs America at least 10 million jobs.
If we run this multiplier by the total number of manufacturing jobs we’ve lost (7 million), then just over 18 million jobs have been displaced by imports.
As you can see, there is no automation job loss: it only appears that way when you ignore the erosion of our output growth due to offshoring.
The Impact Of Automation On Employment Is Neutral
It’s pretty clear that automation doesn’t cause job loss, and you’d think that so many large media outlets would pick up on that fact.
Why are they so incompetent?
Well, it’s not that they’re incompetent (not all of them, anyways), it’s that their interests aren’t aligned with those of ordinary Americans.
The fact is that blaming robots is a convenient smokescreen that masks the elite’s economic shenanigans—if they blame chronic job loss, and the inevitable wage stagnation, on automation or technology, then opposition for economic globalism evaporates.
Well, not all of it. Read my book America Betrayed and support the cause.
Bureau of Labor Statistics. “Civilian labor force participation rate by age, gender, race, and ethnicity.” Accessed June 5, 2016. http://www.bls.gov/emp/ep_table_303.htm
Federal Reserve Bank of St. Louis, “All Employees, Manufacturing.” Accessed Nov 20, 2016. https://fred.stlouisfed.org/series/MANEMP
Maddison, Angus. The World Economy: Historical Statistics. Paris, OECD Publishing, 2003.
Muro, Mark, et al. “America’s Advanced Industries: what are they, where are they, and why they matter.” Brooking’s Institute, 2015.
Nosbuch, Keith D. and John A Bernaden. “The Multiplier Effect.” Manufacturing Executive Leadership Journal (2012).
Scott, Robert E. “Manufacturing Job Loss: trade, not productivity, is the culprit.” Economic Policy Institute Report, 2015.
United States Census Bureau, “Trade in Goods, 1985-2016.” Accessed May 20, 2016. https://www.census.gov/foreign-trade/balance/c5700.html
World Bank, “GDP by PPP Statistics.” Accessed May 15. thttps://knoema.com/mhrzolg/gdp-statistics-from-the-world-bank?country=United%20States